Group Product Manager, AI Security and Privacy Platforms, DeepMind

Google•Mountain View, CA

1d•$240,000 - $333,000

About The Position

This is an opportunity to architect the multi-quarter product goal, organizational strategy, and platform portfolio for AI Security and Privacy Platforms across frontier models, autonomous agent frameworks, evaluation infrastructure, and confidential compute ecosystems. As AI systems evolve into autonomous agents capable of multi-step tool use, code execution, and persistent cross-device context, the role focuses on resilient containment, runtime oversight, and hardware-verifiable privacy to maintain trust and support safe AI development. The Group Product Manager will lead a high-leverage product roadmap embedding agentic security and privacy-preserving architectures across the AI lifecycle, from reinforcement learning and evaluation environments to runtime trajectory monitoring, context lineage tracking, and confidential cloud serving. The role establishes product excellence at the convergence of frontier AI, security and privacy, and governance.

Requirements

Bachelor's degree in Computer Science, Engineering, a related technical discipline, or equivalent practical experience.
10 years of product management experience.
Experience leading platform or infrastructure products in AI/agentic security, confidential computing, systems security, cybersecurity, or privacy engineering contexts.

Nice To Haves

Experience deploying real-time safety guardrails and verifiable security and privacy controls across cloud, air-gapped, or confidential compute environments.
Exceptional executive communication and stakeholder alignment skills, with a proven ability to drive cross-functional alignment across technical and non-technical executive leadership.
Proven ability to steer multi-quarter strategic portfolios bridging foundational research bets with enterprise-grade and consumer-scale production deployments.
Deep technical fluency in agentic security paradigms such as sandbox hardening, runtime reasoning and action monitoring, prompt injection resilience, dynamic information-flow tracking, or automated red-teaming and remediation workflows.

Responsibilities

Define and own the multi-quarter product goal, charter, and roadmap for Agentic Security and Privacy Platforms across frontier training and evaluation pipelines, developer tools, and end-user deployments.
Build secure execution environments across evaluation, reinforcement learning, and developer command-line interface workflows by enforcing hardened sandboxes, strict network access controls, and least-privilege permissions.
Lead unified runtime monitoring platforms combining reasoning-trace oversight, prompt injection defenses, and fine-grained context lineage tracking.
Scale automated adversarial testing platforms to systematically stress-test agent behaviors, paired with verification and safe production deployment.
Lead the platform roadmap for hardware-enforced confidential AI inference—including secure hardware enclaves, remote attestation, and encrypted transport—alongside privacy and safety guardrails for enterprise and sovereign clouds.